Hello Doug,
re: EXDS purchasing GBLX Bandwidth
It makes good sense for any of the larger international SPs with
multinational customers to be acquiring large capacities of international
bandwidth, especially if they have (like EXDS happens to have)
international POPs and Data Centers (IDCs) overseas. Long term
purchases tend to cost only a small fraction of the month to month rates
that they would be paying otherwise. ABOV and ISLD have each done,
or will do, likewise, and FRO will [maybe by default] do the same, as will
all large SPs going forward if they wish to compete globally.
They will need this bandwidth not only for the delivery of end user data
payloads to their client locales, but also for the purposes of improving
latency performance, as well as synchronizing and dual-feeding their own
repositories at some point for added reliability and disaster recovery/
backup purposes, around the globe.
Several aspects of this deal are intriguing to me, however. That EXDS
would directly lease this much bandwidth, all at once, poses some
interesting possibilities.
Does this signal that they are expanding the number of roles they will be
playing from that of a hosting/web-serving service, to that of a full fledged
backbone wholesale provider, say? Or are they simply growing their core
business fast enough to warrant this much additional bandwidth?
Of course, if they are successful in serving even their original corporate
clients who depend on them for data center operations, then it is only a
matter of natural extension that those same corporate clients would begin
looking to EXDS for single sourcing of transport services, as well. Where
does it all end? In the foreseeable future, it doesn't. This could eventually
extend into areas of internet voice, video, etc., far beyond their original
charter.
---------
The Arithmetic
I happened to be glancing at this release earlier. I was examining the
arithmetic behind the bandwidth claims and found it to be more than just a
little bit fuzzy. The "additional" bandwidth [does this imply that they
already had bandwidth from GBLX? if so how much did they previously
have?] of 5.1 Gb/s would equate to some odd number of OC-ns...
roughly 96 T-3s or 96 OC-1's, which would equate to 32 OC-3s.
An OC-3 is rated at 3 T-3s, or 155 Mb/s, and equates to a STM-1.
So, what we are talking about here is a total of 96 T3s, or OC-1s, or 32 OC-3s, which is the same as 32 STM-1s, or ~ 5 Gb/s.
Bandwidth accounting knows no rules, it seems, and aggregated capacities
on ring and meshed networks can be made to look like just about anything
that the marketecture calls for. Take a look at the wording from the
release. Note, the bolded inserts are mine.
"Global Crossing will supply Exodus with two STM-1 [2
OC-3's, or roughly 311 Mb/s] circuits connecting New York to
Frankfurt and two STM-1 [two more OC-3s... thus far we have
an OC-12 worth, equaling 622 Mb/s, sub-totaled] circuits
connecting New York to London; an STM-1 [another OC-3
equal to 155 Mb/s, bringing the total to (622 + 155) = 777Mb/s]
RING connecting London, Amsterdam, Frankfurt, and Paris; four
STM-1 circuits [another OC-12 or 622 Mb/s added to the
already 777 Mb/s equaling 1.399 Gb/s thus far] connecting Santa
Clara and Tokyo; and OC12c [another 622 Mb/s, bringing the
new total to 2.021 Gb/s in total, say 2 Gb/s rounded] service
connecting Los Angeles, Austin, Dallas, Atlanta, and Washington D.C."
Now, how do they come up with 5.Gb/s? I suppose that they are taking
each segment of each ring and individually adding the capacity of those
segments between cities in a linear fashion, as opposed to calling the entire
ring a constant, say, OC-12.
For example, where they claim that they are providing
"an STM-1 ring connecting London, Amsterdam, Frankfurt,
and Paris..." , then that RING should be valued at the STM-1 rate, or a
ring based on 155 Mb/s.
If, on the other hand, they spin it such that each ring segment (London to
Amsterdam being one, Amsterdam to Frankfurt being the second, and so
on) is valued at 155 Mb/s, then there are four segments that could be
stated at 155 Mb/s each, totaling four STM-1s for the entire ring,
or 622 Mb/s, instead of the original 155 Mb/s.
The same could hold true for the North American points, thus doubling,
or quadrupling the stated values of those, as well. It's all in the wording. I
suspect that these accounting techniques are what account for the
apparent discrepancies between what they are calling 5.1 Gb/s and what I
come up with as 2 Gb/s. It's possible that the company could reply
that they are also including backup capacity on the rings, but since that
backup ring capacity is not usable during normal working hours, I would
not count it in the mix. And, of course, it is also possible that there
is an error in my analysis, or that they have not shown all there is to
be shown. I therefore welcome all comments and corrections.
Regards, Frank Coluccio
|
